The Effectiveness of Data Augmentation in Image Classification using Deep Learning

Abstract

In this paper, we explore and compare multiple solutions to the problem of data augmentation in image classification. Previous work has demonstrated the effectiveness of data augmentation through simple techniques, such as cropping, rotating, and flipping input images. We artificially constrain our access to data to a small subset of the ImageNet dataset, and compare each data augmentation technique in turn. One of the more successful data augmentations strategies is the traditional transformations mentioned above. We also experiment with GANs to generate images of different styles. Finally, we propose a method to allow a neural net to learn augmentations that best improve the classifier, which we call neural augmentation. We discuss the successes and shortcomings of this method on various datasets.

8 pages, 12 figures

References in corpus (2)

• Deep Big Simple Neural Nets Excel on Handwritten Digit Recognition
• Recurrent Topic-Transition GAN for Visual Paragraph Generation

Cited by in corpus (262)

• Deep Neural Networks and Tabular Data: A Survey
• Convolutional Neural Networks for Classification of Alzheimer's Disease: Overview and Reproducible Evaluation
• AutoAugment: Learning Augmentation Policies from Data
• A review: Deep learning for medical image segmentation using multi-modality fusion
• Regularized Deep Networks in Intelligent Transportation Systems: A Taxonomy and a Case Study
• A survey on modern trainable activation functions
• Hyper-Parameter Optimization: A Review of Algorithms and Applications
• Improving performance of CNN to predict likelihood of COVID-19 using chest X-ray images with preprocessing algorithms
• Extracting possibly representative COVID-19 Biomarkers from X-Ray images with Deep Learning approach and image data related to Pulmonary Diseases
• Models Genesis
• BAGAN: Data Augmentation with Balancing GAN
• Hydra: an Ensemble of Convolutional Neural Networks for Geospatial Land Classification
• TACTO: A Fast, Flexible, and Open-source Simulator for High-Resolution Vision-based Tactile Sensors
• MIScnn: A Framework for Medical Image Segmentation with Convolutional Neural Networks and Deep Learning
• Selective Synthetic Augmentation with HistoGAN for Improved Histopathology Image Classification
• Deep Learning in Mobile and Wireless Networking: A Survey
• Further advantages of data augmentation on convolutional neural networks
• Dynamically enhanced static handwriting representation for Parkinson's disease detection
• Cross-Domain Few-Shot Classification via Learned Feature-Wise Transformation
• Recent Advances in Object Detection in the Age of Deep Convolutional Neural Networks
• Combining Many-objective Radiomics and 3-dimensional Convolutional Neural Network through Evidential Reasoning to Predict Lymph Node Metastasis in Head and Neck Cancer
• Image to Images Translation for Multi-Task Organ Segmentation and Bone Suppression in Chest X-Ray Radiography
• A physics-informed operator regression framework for extracting data-driven continuum models
• Generative Data Augmentation for Commonsense Reasoning
• Study of Different Deep Learning Approach with Explainable AI for Screening Patients with COVID-19 Symptoms: Using CT Scan and Chest X-ray Image Dataset
• Label-Only Membership Inference Attacks
• Data augmentation instead of explicit regularization
• Deep Reinforcement Learning Optimizes Graphene Nanopores for Efficient Desalination
• Adversarial Attacks Against Deep Generative Models on Data: A Survey
• Wireframe-Based UI Design Search Through Image Autoencoder
• Deep Long-Tailed Learning: A Survey
• Local Motion Planner for Autonomous Navigation in Vineyards with a RGB-D Camera-Based Algorithm and Deep Learning Synergy
• Improving Malaria Parasite Detection from Red Blood Cell using Deep Convolutional Neural Networks
• Deep Learning vs. Traditional Computer Vision
• Integrating uncertainty in deep neural networks for MRI based stroke analysis
• A Survey of Active Learning for Text Classification using Deep Neural Networks
• Co-Correcting: Noise-tolerant Medical Image Classification via mutual Label Correction
• On Leveraging Pretrained GANs for Generation with Limited Data
• OSLNet: Deep Small-Sample Classification with an Orthogonal Softmax Layer
• Adam revisited: a weighted past gradients perspective
• Generative Image Translation for Data Augmentation in Colorectal Histopathology Images
• Machine learning astrophysics from 21 cm lightcones: impact of network architectures and signal contamination
• Skin Lesions Classification Using Convolutional Neural Networks in Clinical Images
• Fine-tuning deep learning model parameters for improved super-resolution of dynamic MRI with prior-knowledge
• Two Decades of Bengali Handwritten Digit Recognition: A Survey
• Neural Networks Versus Conventional Filters for Inertial-Sensor-based Attitude Estimation
• A Multi-cascaded Model with Data Augmentation for Enhanced Paraphrase Detection in Short Texts
• GaNDLF: A Generally Nuanced Deep Learning Framework for Scalable End-to-End Clinical Workflows in Medical Imaging
• Using Machine Learning Safely in Automotive Software: An Assessment and Adaption of Software Process Requirements in ISO 26262
• Neural Data Augmentation via Example Extrapolation
• Deep learning approach for identification of HII regions during reionization in 21-cm observations
• Contrastive Learning with Stronger Augmentations
• A Causal View on Robustness of Neural Networks
• Theoretical evidence for adversarial robustness through randomization
• Deep Weakly-supervised Anomaly Detection
• When Deep Learning Meets Multi-Task Learning in SAR ATR: Simultaneous Target Recognition and Segmentation
• Improving Sequence-to-Sequence Acoustic Modeling by Adding Text-Supervision
• Augmentation Scheme for Dealing with Imbalanced Network Traffic Classification Using Deep Learning
• How Good is the Bayes Posterior in Deep Neural Networks Really?
• Uncertainty Quantification and Deep Ensembles
• Anatomy-aware and acquisition-agnostic joint registration with SynthMorph
• An ordinal CNN approach for the assessment of neurological damage in Parkinson's disease patients
• Decreasing False Alarm Rates in ML-based Solar Flare Prediction using SDO/HMI Data
• Improving Deep Learning Models via Constraint-Based Domain Knowledge: a Brief Survey
• MixText: Linguistically-Informed Interpolation of Hidden Space for Semi-Supervised Text Classification
• A Four-Stage Data Augmentation Approach to ResNet-Conformer Based Acoustic Modeling for Sound Event Localization and Detection
• Mixing Real and Synthetic Data to Enhance Neural Network Training -- A Review of Current Approaches
• Viewmaker Networks: Learning Views for Unsupervised Representation Learning
• CAD2Render: A Modular Toolkit for GPU-accelerated Photorealistic Synthetic Data Generation for the Manufacturing Industry
• Selecting Data Augmentation for Simulating Interventions
• GraphCrop: Subgraph Cropping for Graph Classification
• Automatic detection of microsleep episodes with deep learning
• Photometric identification of compact galaxies, stars and quasars using multiple neural networks
• Robust deep learning for eye fundus images: Bridging real and synthetic data for enhancing generalization
• Self-supervised Learning for Panoptic Segmentation of Multiple Fruit Flower Species
• Deep-CEE I: Fishing for Galaxy Clusters with Deep Neural Nets
• Technical Considerations for Semantic Segmentation in MRI using Convolutional Neural Networks
• Feature Re-Learning with Data Augmentation for Video Relevance Prediction
• Model Assertions for Monitoring and Improving ML Models
• Real-time Federated Evolutionary Neural Architecture Search
• DARE: Data Augmented Relation Extraction with GPT-2
• On Data-Augmentation and Consistency-Based Semi-Supervised Learning
• Classification of Multiwavelength Transients with Machine Learning
• Neural Sign Language Translation based on Human Keypoint Estimation
• Synergy between Machine/Deep Learning and Software Engineering: How Far Are We?
• BdSL36: A Dataset for Bangladeshi Sign Letters Recognition
• Sequential IoT Data Augmentation using Generative Adversarial Networks
• Exploring Bias in GAN-based Data Augmentation for Small Samples
• Active Image Synthesis for Efficient Labeling
• Orchestrating the Development Lifecycle of Machine Learning-Based IoT Applications: A Taxonomy and Survey
• TDEFSI: Theory Guided Deep Learning Based Epidemic Forecasting with Synthetic Information
• Syntactic Data Augmentation Increases Robustness to Inference Heuristics
• Analysis of Classifier Training on Synthetic Data for Cross-Domain Datasets
• Mine Your Own vieW: Self-Supervised Learning Through Across-Sample Prediction
• Influence-guided Data Augmentation for Neural Tensor Completion
• EllSeg-Gen, towards Domain Generalization for head-mounted eyetracking
• Deep learning pipeline for image classification on mobile phones
• Efficient ConvNets for Analog Arrays
• A Deep Learning Approach for Determining Effects of Tuta Absoluta in Tomato Plants
• Counterfactual Data Augmentation using Locally Factored Dynamics
• Using generative adversarial networks to synthesize artificial financial datasets
• Contrastive-mixup learning for improved speaker verification
• SCARF: Self-Supervised Contrastive Learning using Random Feature Corruption
• Alleviating Mode Collapse in GAN via Diversity Penalty Module
• Building Robust Deep Neural Networks for Road Sign Detection
• Fuzzy Unique Image Transformation: Defense Against Adversarial Attacks On Deep COVID-19 Models
• Tractable Regularization of Probabilistic Circuits
• Morphological Classification of Galaxies Using SpinalNet
• Abnormal Respiratory Sound Identification Using Audio-Spectrogram Vision Transformer
• An Analysis of Simple Data Augmentation for Named Entity Recognition
• Online Hyper-parameter Learning for Auto-Augmentation Strategy
• Data Augmentation Using Adversarial Training for Construction-Equipment Classification
• SIP-SegNet: A Deep Convolutional Encoder-Decoder Network for Joint Semantic Segmentation and Extraction of Sclera, Iris and Pupil based on Periocular Region Suppression
• Scale-aware Automatic Augmentation for Object Detection
• Graph-based Interpolation of Feature Vectors for Accurate Few-Shot Classification
• Use of Transfer Learning and Wavelet Transform for Breast Cancer Detection
• Data Augmentation for Object Detection via Differentiable Neural Rendering
• Automated Image Data Preprocessing with Deep Reinforcement Learning
• MSE Loss with Outlying Label for Imbalanced Classification
• Cycle-Consistent Adversarial Autoencoders for Unsupervised Text Style Transfer
• Domain adaptation for holistic skin detection
• Implicit Rugosity Regularization via Data Augmentation
• Virtual Underwater Datasets for Autonomous Inspections
• On Data Augmentation and Adversarial Risk: An Empirical Analysis
• Robustness of Visual Explanations to Common Data Augmentation
• Adversarial Learning of General Transformations for Data Augmentation
• A Deep-learning-based Method for PIR-based Multi-person Localization
• DADA: Deep Adversarial Data Augmentation for Extremely Low Data Regime Classification
• Data Augmentation for Instrument Classification Robust to Audio Effects
• Style Transfer with Time Series: Generating Synthetic Financial Data
• Learn to segment single cells with deep distance estimator and deep cell detector
• CAVA: A Visual Analytics System for Exploratory Columnar Data Augmentation Using Knowledge Graphs
• Exploring Deep Learning Methods for Classification of SAR Images: Towards NextGen Convolutions via Transformers
• A Self-Supervised Feature Map Augmentation (FMA) Loss and Combined Augmentations Finetuning to Efficiently Improve the Robustness of CNNs
• Automated Copper Alloy Grain Size Evaluation Using a Deep-learning CNN
• More Than Meets The Eye: Semi-supervised Learning Under Non-IID Data
• CLCC: Contrastive Learning for Color Constancy
• ObjectAug: Object-level Data Augmentation for Semantic Image Segmentation
• Deep learning in bioinformatics: introduction, application, and perspective in big data era
• SeismoGen: Seismic Waveform Synthesis Using Generative Adversarial Networks
• Human-Expert-Level Brain Tumor Detection Using Deep Learning with Data Distillation and Augmentation
• A Meta Approach to Defend Noisy Labels by the Manifold Regularizer PSDR
• Synthetic Sample Selection via Reinforcement Learning
• High-contrast "gaudy" images improve the training of deep neural network models of visual cortex
• Accurate pedestrian localization in overhead depth images via Height-Augmented HOG
• CIS-Net: A Novel CNN Model for Spatial Image Steganalysis via Cover Image Suppression
• Selective Synthetic Augmentation with Quality Assurance
• Dynamic Stale Synchronous Parallel Distributed Training for Deep Learning
• Snore-GANs: Improving Automatic Snore Sound Classification with Synthesized Data
• Learning Optimal Data Augmentation Policies via Bayesian Optimization for Image Classification Tasks
• Inducing Causal Structure for Interpretable Neural Networks
• Melodic Phrase Segmentation By Deep Neural Networks
• A two-step machine learning approach for crop disease detection: an application of GAN and UAV technology
• On the Pitfalls of Learning with Limited Data: A Facial Expression Recognition Case Study
• MRFusion: A Deep Learning architecture to fuse PAN and MS imagery for land cover mapping
• Data Augmentation for Enhancing EEG-based Emotion Recognition with Deep Generative Models
• Generative Interventions for Causal Learning
• Style transfer-based image synthesis as an efficient regularization technique in deep learning
• DeltaGAN: Towards Diverse Few-shot Image Generation with Sample-Specific Delta
• SimEx: Express Prediction of Inter-dataset Similarity by a Fleet of Autoencoders
• Distinguishing rule- and exemplar-based generalization in learning systems
• Reducing and Exploiting Data Augmentation Noise through Meta Reweighting Contrastive Learning for Text Classification
• Data augmentation and image understanding
• EEG Classification by factoring in Sensor Configuration
• A Survey of Techniques All Classifiers Can Learn from Deep Networks: Models, Optimizations, and Regularization
• ECG-DelNet: Delineation of Ambulatory Electrocardiograms with Mixed Quality Labeling Using Neural Networks
• GAN Memory with No Forgetting
• Text Data Augmentation: Towards better detection of spear-phishing emails
• Improving the Performance of Fine-Grain Image Classifiers via Generative Data Augmentation
• Learning Invariances for Policy Generalization
• Revisiting data augmentation for subspace clustering
• Class balanced underwater object detection dataset generated by class-wise style augmentation
• Intelligence plays dice: Stochasticity is essential for machine learning
• Performance of a Geometric Deep Learning Pipeline for HL-LHC Particle Tracking
• CNN Feature Map Augmentation for Single-Source Domain Generalization
• Automatic Data Augmentation by Learning the Deterministic Policy
• Effective Evaluation of Deep Active Learning on Image Classification Tasks
• Landmarks Augmentation with Manifold-Barycentric Oversampling
• The Effect of Visual Design in Image Classification
• Evaluating Self and Semi-Supervised Methods for Remote Sensing Segmentation Tasks
• Automatically Learning Data Augmentation Policies for Dialogue Tasks
• "I have vxxx bxx connexxxn!": Facing Packet Loss in Deep Speech Emotion Recognition
• Malware Classification Using Deep Boosted Learning
• Using Ensemble Classifiers to Detect Incipient Anomalies
• RIANN -- A Robust Neural Network Outperforms Attitude Estimation Filters
• Learning Vine Copula Models For Synthetic Data Generation
• BERT for Large-scale Video Segment Classification with Test-time Augmentation
• General Probabilistic Surface Optimization and Log Density Estimation
• Style-invariant Cardiac Image Segmentation with Test-time Augmentation
• On the effectiveness of convolutional autoencoders on image-based personalized recommender systems
• Seq2Seq and Multi-Task Learning for joint intent and content extraction for domain specific interpreters
• Data augmentation with Mobius transformations
• Towards Feature-free TSP Solver Selection: A Deep Learning Approach
• Lesion Conditional Image Generation for Improved Segmentation of Intracranial Hemorrhage from CT Images
• Multiscale Attention Guided Network for COVID-19 Diagnosis Using Chest X-ray Images
• Automated Annotation of Scientific Texts for ML-based Keyphrase Extraction and Validation
• Thinking Outside the Pool: Active Training Image Creation for Relative Attributes
• Calibration and Uncertainty for multiRater Volume Assessment in multiorgan Segmentation (CURVAS) challenge results
• Artificial Intelligence in Quantitative Ultrasound Imaging: A Review
• LADA: Look-Ahead Data Acquisition via Augmentation for Active Learning
• Data Augmentation for Text Generation Without Any Augmented Data
• Distinguishing artefacts: evaluating the saturation point of convolutional neural networks
• Learning Fast and Precise Pixel-to-Torque Control
• PSF-based Analysis for Detecting Unresolved Wide Binaries
• DNN-based workflow for attenuating seismic interference noise and its application to marine towed streamer data from the Northern Viking Graben
• Void region segmentation in ball grid array using u-net approach and synthetic data
• Liver Segmentation using Turbolift Learning for CT and Cone-beam C-arm Perfusion Imaging
• What augmentations are sensitive to hyper-parameters and why?
• Frequency Disentangled Residual Network
• The Tags Are Alright: Robust Large-Scale RFID Clone Detection Through Federated Data-Augmented Radio Fingerprinting
• Dynamic Defense Approach for Adversarial Robustness in Deep Neural Networks via Stochastic Ensemble Smoothed Model
• Automatic detection of impact craters on Al foils from the Stardust interstellar dust collector using convolutional neural networks
• Data Augmentation for Deep Candlestick Learner
• LSTM and GPT-2 Synthetic Speech Transfer Learning for Speaker Recognition to Overcome Data Scarcity
• Improving Text Relationship Modeling with Artificial Data
• Discriminative Cross-Modal Data Augmentation for Medical Imaging Applications
• Towards Leveraging End-of-Life Tools as an Asset: Value Co-Creation based on Deep Learning in the Machining Industry
• Black-box Adversarial Sample Generation Based on Differential Evolution
• Balanced Order Batching with Task-Oriented Graph Clustering
• Embedded Emotions -- A Data Driven Approach to Learn Transferable Feature Representations from Raw Speech Input for Emotion Recognition
• 3D-Aided Data Augmentation for Robust Face Understanding
• A Generative Approach Towards Improved Robotic Detection of Marine Litter
• Cross-Domain Image Classification through Neural-Style Transfer Data Augmentation
• Simulation of virtual cohorts increases predictive accuracy of cognitive decline in MCI subjects
• Improved visible to IR image transformation using synthetic data augmentation with cycle-consistent adversarial networks
• Building robust prediction models for defective sensor data using Artificial Neural Networks
• Self-paced Data Augmentation for Training Neural Networks
• Building Compact and Robust Deep Neural Networks with Toeplitz Matrices
• Automatic joint damage quantification using computer vision and deep learning
• Trainable back-propagated functional transfer matrices
• What Else Can Fool Deep Learning? Addressing Color Constancy Errors on Deep Neural Network Performance
• Spectral Data Augmentation Techniques to quantify Lung Pathology from CT-images
• Class Interference Regularization
• Crowding in humans is unlike that in convolutional neural networks
• Robustness and Overfitting Behavior of Implicit Background Models
• Tangent Space Sensitivity and Distribution of Linear Regions in ReLU Networks
• Achieving Model Robustness through Discrete Adversarial Training
• Sinusoidal wave generating network based on adversarial learning and its application: synthesizing frog sounds for data augmentation
• Data augmentation using generative networks to identify dementia
• Classification of animal sounds in a hyperdiverse rainforest using Convolutional Neural Networks
• Handcrafted vs Deep Learning Classification for Scalable Video QoE Modeling
• Detecting cities in aerial night-time images by learning structural invariants using single reference augmentation
• WeedMap: A large-scale semantic weed mapping framework using aerial multispectral imaging and deep neural network for precision farming
• Fast Jacobian-Vector Product for Deep Networks
• A novel multiple instance learning framework for COVID-19 severity assessment via data augmentation and self-supervised learning
• Context Normalization Layer with Applications
• Label Denoising with Large Ensembles of Heterogeneous Neural Networks
• Tips and Tricks for Webly-Supervised Fine-Grained Recognition: Learning from the WebFG 2020 Challenge
• Jitter: Random Jittering Loss Function
• GAN based ball screw drive picture database enlargement for failure classification
• Efficient Modelling Across Time of Human Actions and Interactions
• On tuning deep learning models: a data mining perspective
• Equivariance-bridged SO(2)-Invariant Representation Learning using Graph Convolutional Network
• Pathology-Aware Generative Adversarial Networks for Medical Image Augmentation
• Where is the disease? Semi-supervised pseudo-normality synthesis from an abnormal image
• Unsupervised Image Segmentation by Mutual Information Maximization and Adversarial Regularization
• Nearest Neighbour Few-Shot Learning for Cross-lingual Classification
• Safe Augmentation: Learning Task-Specific Transformations from Data
• Implementation of Optical Deep Neural Networks using the Fabry-Perot Interferometer
• An Abstraction Model for Semantic Segmentation Algorithms
• Regularising Deep Networks with Deep Generative Models
• A Neural Network Based Framework for Archetypical Sound Synthesis